This document summarizes a psychological study that examined how adult attachment orientations can systematically bias memories of relationship events. The study had couples engage in videotaped conflict discussions. Each partner then rated their own behavior immediately after and again one week later. Results showed that more avoidant individuals remembered being less supportive if distressed, whereas less avoidant individuals remembered being more supportive if distressed. More anxious individuals remembered being less distant if distressed, whereas less anxious individuals remembered being more distant if distressed. The study provides evidence that attachment orientations can twist memories in systematic ways.

1. Psychological Science http://pss.sagepub.com/
Attachment Working Models Twist Memories of Relationship Events
Jeffry A. Simpson, W. Steven Rholes and Heike A. Winterheld
Psychological Science 2010 21: 252 originally published online 22 December 2009
DOI: 10.1177/0956797609357175
The online version of this article can be found at:
http://pss.sagepub.com/content/21/2/252
Published by:
http://www.sagepublications.com
On behalf of:
Association for Psychological Science
Additional services and information for Psychological Science can be found at:
Email Alerts: http://pss.sagepub.com/cgi/alerts
Subscriptions: http://pss.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
>> Version of Record - Feb 18, 2010
OnlineFirst Version of Record - Dec 22, 2009
What is This?
Downloaded from pss.sagepub.com by guest on January 29, 2013

2. Research Article
Psychological Science
Attachment Working Models Twist Memories 21(2) 252­ 259
–
© The Author(s) 2010
Reprints and permission: http://www
of Relationship Events .sagepub.com/journalsPermissions.nav
DOI: 10.1177/0956797609357175
http://pss.sagepub.com
Jeffry A. Simpson1, W. Steven Rholes2, and Heike A. Winterheld3
1
University of Minnesota, Twin Cities Campus; 2Texas A&M University; and 3California State University, East Bay
Abstract
The information that people remember about their relationships should be affected by their attachment orientations. This
study investigated changes in individuals’ memories of their own behavior during conflict-resolution discussions with their
romantic partners. One week after each couple engaged in two videotaped conflict discussions, each partner’s memory of the
discussions was assessed. Memory biases were systematically related to attachment orientations. More avoidant individuals, for
example, remembered being less supportive than they reported initially if they were relatively distressed during the discussions,
whereas the opposite was true of less avoidant persons. More anxious individuals remembered being less emotionally distant
than they reported initially if they were relatively distressed during the discussions, whereas the opposite was true of less
anxious persons. We discuss the implications of these memory changes.
Keywords
attachment, working models, memory, conflict resolution
Received 12/4/08; Revision accepted 6/1/09
According to attachment theory (Bowlby, 1973, 1980), experi- who score high on one or both of these dimensions are inse-
ences with caregivers and other attachment figures during curely attached, whereas those who score low on both dimen-
development create mental representations, or working mod- sions are securely attached. Highly secure individuals are
els, of the self and others. These models encompass autobio- comfortable with closeness, are open to depending on others
graphical memories of specific experiences with attachment and having others depend on them, and do not worry about
figures, beliefs and attitudes about attachment figures, and being abandoned.
conditional “if/then” rules (e.g., “if I am upset, then my part- Attachment orientations and the working models that
ner will/will not support me”) that result in expectations underlie them affect how people process interpersonal infor-
of partners and relationships. These models also contain mation in numerous important ways (Bowlby, 1980; Collins
procedural-knowledge rules about how to regulate emotions et al., 2004). Several studies have examined how relationship-
and thoughts in order to achieve attachment-related goals relevant perceptions and judgments are shaped by attachment
(Collins, Guichard, Ford, & Feeney, 2004; Mikulincer & orientations (e.g., Belsky, Spritz, & Crnic, 1996; Fraley, Gar-
Shaver, 2003). Over time, working models produce unique ner, & Shaver, 2000; Kirsh & Cassidy, 1997; Mikulincer &
patterns of affect, cognition, and behavior that reflect different Arad, 1999; Mikulincer & Orbach, 1995; Miller, 1999; Miller
adult attachment orientations. & Noirot, 1999; Pietromonaco & Barrett, 1997; Rudolph,
Adult attachment orientations are measured on two orthog- Hammen, & Burge, 1997). Collectively, these studies have
onal dimensions (Brennan, Clark, & Shaver, 1998; Simpson, found that highly avoidant and highly anxious people perceive
Rholes, & Phillips, 1996), known as avoidance and anxiety. their social environments more negatively than do securely
Individuals who score high on the avoidance dimension desire attached people, and that insecure people tend to perceive neg-
psychological autonomy, independence, and sufficient emo- ative relationship events more accurately than positive ones
tional distance from their partners; these desires motivate (see also Collins & Feeney, 2004; Kirsh & Cassidy, 1997).
avoidant individuals to limit and control caregiving, intimacy,
and interdependence. Individuals who score high on the anxi-
Corresponding Author:
ety dimension crave psychological closeness and intimacy Jeffry A. Simpson, Department of Psychology, University of Minnesota,
with their partners, but worry that their partners may be unable Minneapolis, MN 55455-0344
or unwilling to provide sufficient care and support. Individuals E-mail: simps108@umn.edu
Downloaded from pss.sagepub.com by guest on January 29, 2013

3. Attachment and Memory 253
The effects that attachment orientations have on memory activated and guide how people think, feel, and behave princi-
have seldom been studied. One of the few studies that exam- pally when they are distressed. The effects of distress on observ-
ined such effects (B.C. Feeney & Cassidy, 2003) investigated able behavior have been studied extensively. This research has
how adolescents’ attachment-related representations of each confirmed that attachment orientations affect behavior most
parent predicted memories of conflict interactions with each strongly when individuals feel threatened, overly challenged, or
parent 6 weeks earlier. The study found that the memories of distressed (J.A. Feeney, 1999; Kobak & Duemmler, 1994; Miku-
adolescents who reported less secure representations of their lincer, Florian, & Weller, 1993; Simpson, Rholes, & Nelligan,
parents became more negative over time than did the memo- 1992). These findings support Bowlby’s (1969) claim that one
ries of adolescents who had more secure representations. major function of the attachment system is to prepare individuals
Memory-change effects such as these are important because to respond to and deal with threatening or distressing events. In
schema-consistent memories are easily assimilated into, and the absence of threat or stress, however, other behavioral systems
may help to maintain, the working models that underlie adult (e.g., exploration) should govern behavior.
attachment orientations (Fraley & Brumbaugh, 2004). The effects of distress and adult attachment orientations on
In the research we report here, we examined how romantic memory for attachment-relevant information have never been
attachment orientations systematically twist individuals’ mem- examined. We anticipated, however, that distress would have
ory for their own behavior in a videotaped attachment-relevant the same kinds of effects on memory that it does on overt
situation. To date, no studies of memory in the attachment lit- behavior, given that working models become more salient
erature have examined systematic changes in self-relevant when individuals are distressed.
memories. This lack of research represents a significant gap in
knowledge because autobiographical memories and associated
self-concepts are likely to affect the way in which individuals Method
think, feel, and behave in close relationships. Participants
We first measured the attachment orientations of both part-
ners in romantic relationships. We then had each couple Participants were 148 heterosexual couples from a Southwest-
engage in a videotaped conflict-resolution discussion task. ern university. Mean relationship length was 19.93 months
This task created an attachment-relevant event that both part- (SD = 16.36 months; range = 3–108 months). Eighty-nine per-
ners would later be asked to recall. Immediately after each dis- cent of the participants were in dating relationships, 8% were
cussion, each partner rated how supportive and emotionally engaged, and 3% were married. The mean ages of the men and
distant he or she had been during the discussion. The con- women were 19.72 and 19.00 years, respectively (ranges =
structs of support and emotional distance are both central to 18–30 for men and 18–30 for women). Eighty percent were
attachment theory (Bowlby, 1973, 1980). One week later, each Caucasian, 12% were Hispanic, 5% were Asian American, 1%
partner completed the same support and distance measures were African American, and 2% were multiracial.
from memory. This allowed us to test the degree to which each
partner’s memory of his or her behavior differed from his or
her immediate postdiscussion perceptions. Phase 1: Procedure and measures
We predicted that avoidant individuals would remember being Questionnaires. When couples arrived for Phase 1, each part-
less supportive and more distancing than indicated by their per- ner first completed questionnaires in a private room. Embed-
ceptions immediately following the conflict-discussion task, ded in the questionnaires were demographic questions along
given their needs and goals for achieving and maintaining psy- with (a) the Adult Attachment Questionnaire (AAQ; Simpson
chological and emotional distance. We also predicted that anxious et al., 1996), a well-validated 17-item measure that assesses
individuals would recall being less distancing and more suppor­ attachment anxiety (e.g., “I usually want more closeness and
tive than indicated by their immediate postdiscussion perceptions, intimacy than others do”) and avoidance (e.g., “Others often
given their needs and goals for attaining greater closeness and felt want me to be more intimate than I feel comfortable being”),
security. These predicted “memory twists” would be consistent and (b) the Berkeley Personality Profile (John, Donahue, &
with the chronic needs, concerns, and interpersonal goals har- Kentle, 1991), which was included to assess neuroticism for
bored by highly avoidant and highly anxious people (see Miku- discriminant-validity purposes (see Brennan & Shaver, 1995).
lincer & Shaver, 2003). Moreover, we expected that these effects Alphas for these scales ranged from .72 to .80.
would reflect actual memory twists and therefore would remain
significant even when we statistically controlled for observer rat- Conflict discussion task. After completing the questionnaires,
ings of each individual’s supportiveness and distancing. both partners were led to the same room. The experimenter
We also reasoned, however, that these effects might be mod- then asked each partner to list the two most serious, unresolved
erated by the degree to which individuals were distressed during problems in the relationship. The two highest-ranked prob-
the conflict-discussion task (Alexander, Quas, & Goodman, lems (one from each partner’s list) were then discussed in
2002; Simpson & Rholes, 1994). According to attachment the- separate videotaped discussions. The experimenter gave each
ory (Bowlby, 1969, 1973), the attachment system should become couple the following instructions for the first discussion:
Downloaded from pss.sagepub.com by guest on January 29, 2013

4. 254 Simpson et al.
Your partner has identified an area of current conflict in create memory scores for supportiveness and distancing
your relationship. We would now like both of you to (recalled supportiveness: α = .91; recalled distancing: α = .74).
think about this issue for a few moments. Think about
what it is that upsets you, why this is an issue of concern Observer ratings. Each discussion was then rated by nine
in your relationship, and how it might be resolved. independent observers who were blind to all hypotheses and
other data. Observers rated the behavior of each partner using the
When the first videotaped discussion was finished (6–8 min), same supportiveness and distancing scales on which participants
the same procedure was used for the second discussion, during had rated themselves. Because the average interrater reliabil-
which the other partner’s highest-ranked conflict issue was ity across all rated items was .81, the ratings for each item
discussed. were averaged across the nine raters. The items defined two
factors: Supportiveness and Distancing. We therefore com-
Perceptions of support and distancing. We assessed part- puted an observer-rated supportiveness score (α = .99) and an
ners’ perceptions of how supportive and distancing they thought observer-rated distancing score (α = .82) for each participant.
they had been immediately after each discussion. Specifically, Observers also rated how distressed or anxious each partner
each participant rated his or her level of support (e.g., “I was appeared during the discussions, using 9-point Likert-type
supportive of my partner,” “I was sympathetic to my partner’s scales. Factor analysis revealed a single Stress-Anxiety dimen-
point of view”) and distancing (e.g., “I was emotionally dis- sion, so a stress-anxiety score (α = .91) was calculated for each
tant,” “I was withdrawn or disengaged during the discussion”) participant.
by rating 10 items adapted from a scale by Cutrona (1989). Each
item was answered on a 9-point Likert-type scale ranging from
1, not at all, to 9, extremely. The 10 items loaded on two factors: Results
Supportiveness and Distancing. Ratings on the appropriate We report the results in three sections. First, we present the
items were averaged to create scores for perceived supportive- results of analyses examining the effects of predictor variables
ness and perceived distancing for each participant (perceived on observer ratings of supportiveness and distancing. Second,
supportiveness: α = .92; perceived distancing: α = .69). we report the results of analyses testing the hypothesized
effects of individuals’ (actors’) attachment orientations and
rated levels of distress; these analyses tested whether attach-
Phase 2: Procedure, measures, ment and distress predicted specific memory twists involving
and behavioral coding recollections of supportiveness and distancing. Finally, we
Memories of supportiveness and distancing. When each report the results of analyses that provide discriminant-­ alidity
v
couple returned for Phase 2, partners were led to separate evidence for the findings. The data were analyzed using dyadic
rooms. Each partner first used a 7-point Likert-type scale to multiple regression techniques that take into account the sta-
report how much the couple had talked about the videotaped tistical dependency that exists between partners in relation-
discussion task during the preceding week. Then, on the same ships (the Actor-Partner Interdependence Model, or APIM;
9-point scales used in Phase 1, each partner rated how sup- Kashy & Kenny, 2000; Kenny, 1996).
portive and distancing he or she remembered being during the Table 1 reports descriptive statistics for the primary vari-
conflict discussions 1 week earlier. Factor analyses again ables in the study and results of t tests of gender differences.
revealed a Supportiveness and a Distancing factor for these Correlations among the primary variables are provided in
ratings. Ratings on the appropriate items were averaged to Table 2.
Table 1. Mean Ratings for Men and Women and Statistical Tests of Gender Differences
Variable Men Women Mean difference Matched-pairs t test
Avoidance 24.55 (7.85) 23.20 (8.75) 1.35 t(147) = 1.45, n.s.
Anxiety 29.26 (8.30) 26.84 (9.53) 2.42 t(147) = 2.41, p < .02
Neuroticism 16.49 (4.61) 21.59 (5.77) –5.10 t(147) = –8.23, p < .001
Self-perceived supportiveness (Phase 1) 52.00 (9.47) 52.51 (10.87) –0.51 t(146) = –0.57, n.s.
Self-perceived distancing (Phase 1) 7.11 (4.53) 7.00 (4.54) 0.11 t(147) = 0.22, n.s.
Observer-rated supportiveness (Phase 1) 38.87 (9.35) 35.22 (9.61) 3.65 t(146) = 4.71, p < .001
Observer-rated distancing (Phase 1) 9.64 (2.75) 9.44 (3.64) 0.20 t(146) = 0.52, n.s.
Change in self-perceived supportiveness (Phase 1 to Phase 2) –1.43 (6.27) –2.02 (6.64) 0.59 t(144) = 0.81, n.s.
Change in self-perceived distancing (Phase 1 to Phase 2) 0.16 (3.89) –0.27 (4.31) 0.44 t(145) = 0.96, n.s.
Observer-rated distress 2.63 (0.85) 3.05 (1.02) –0.42 t(148) = –5.40, p <.001
Note: N = 148 men and 148 women. Standard deviations of the ratings are given in parentheses.
Downloaded from pss.sagepub.com by guest on January 29, 2013

5. Attachment and Memory 255
Table 2. Correlations Among the Variables
Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
1. ale
M — .40** –.04 –.29** .31** –.02 .10 .01 .22** .06 –.01 –.05 .01 .01 .00 .17* .05 .16
avoidance
2. ale
M — .05 –.13 .14 –.05 .07 .03 .12 .17* .06 –.11 .00 .02 –.12 .20* .04 .08
anxiety
3. ale
M — –.28** .10 –.60** .27** –.21* .16 .01 .06 .53** –.36** .02 –.57** .04 –.12 .06
distress
4. ale
M — –.35** .41** –.25** .00 –.24** –.06 –.21** –.25** .40** –.05 .31** –.08 .07 –.11
perceived
support
(P1)
5. ale
M — –.33** .41** –.11 .00 .10 .13 .12 –.21* .11 –.21* .20* .09 .07
perceived
distancing
(P1)
6. ale
M — –.76** .34** –.16 –.08 –.05 –.37** .41** –.02 .53** –.17* .07 –.08
support
7. ale
M — –.16* .09 .07 .04 .17* –.30** .04 –.32** .16 .03 .14
distancing
8. hange
C — –.03 .01 .06 –.14 .14 .00 .12 .02 .11 –.04
in male
perceived
support
9. hange
C — .01 –.05 .04 –.14 –.03 –.23** .19* –.18* .12
in male
perceived
distancing
10. emale
F — .27** –.03 –.12 .26** .15 .27** –.01 –.14
avoidance
11. emale
F — .23** –.10 .18* –.12 .08 .09 .03
anxiety
12. emale
F — –.38** –.06 –.52** .00 –.04 –.03
distress
13. emale
F — –.11 .51** .00 .00 –.04
perceived
support
(P1)
14. emale
F — –.20* .29** –.08 .00
perceived
distancing
(P1)
15. emale
F — –.34** .28** –.05
support
16. emale
F — .18* .20*
distancing
17. hange
C — –.28**
in female
perceived
support
18. hange
C —
in female
perceived
distancing
Note: N = 148 women and 148 men. All correlations are two-tailed. Anxiety and avoidance were measured by the Adult Attachment Questionnaire (AAQ;
Simpson, Rholes, Phillips, 1996). “Perceived support” and “perceived distancing” refer to participants’ self-report on these measures; “distress,” “support,”
and “distancing” refer to observers’ ratings. P1 = Phase 1.
*p .05. **p .01.
Downloaded from pss.sagepub.com by guest on January 29, 2013

6. 256 Simpson et al.
Predicting observer ratings of 3
conflict-discussion behavior Low Actor Avoidance
Change in Perceived Supportiveness
2 High Actor Avoidance
The predictor variables in the first set of APIM analyses were
actor gender, actor and partner avoidance and anxiety, and actor
and partner observer-rated distress. The dependent variables
1
were the observer ratings of each individual’s (each actor’s)
supportiveness and distancing in the discussions. More avoidant
individuals (actors) were rated as behaving in a less supportive 0
and more distancing manner during the discussions, β = –0.12, Low Actor Distress High Actor Distress
t(252) = –2.12, p .04, and β = 0.07, t(261) = 2.88, p .005,
respectively. In addition, less anxious individuals (actors) were −1
rated as behaving more supportively when they had partners
who were rated as less distressed, β = 0.18, t(240) = 2.37, p .02.
The analyses also indicated that individuals (actors) who were −2
rated as more distressed, β = –4.58, t(242) = –8.24, p .001, and
individuals (actors) who had more distressed partners, β =
–2.55, t(242) = –4.63, p .001, were both rated as behaving less −3
supportively. Finally, individuals (actors) who were rated as Fig. 1. The interaction of actor’s distress and attachment avoidance in
more distressed and had more avoidant partners were rated as predicting change in actor’s self-perceptions of supportiveness from Phase 1
to Phase 2. Slopes were computed for values 1 standard deviation above and
behaving less supportively, β = 0.17, t(265) = 1.97, p .05. 1 standard deviation below the mean on each independent variable.
Differences between immediate perceptions
and later recollections partners than they had initially reported if they had been rela-
tively distressed during the discussion, β = –.06, t(258) =
To test for the hypothesized memory changes, we created resid- –2.00, p .05. This interaction also indicated that more anx-
ualized scores in which each self-perception measure (col- ious individuals (actors) remembered being less distant from
lected immediately after each discussion) was partialed from the their partners (i.e., closer to their partners) than they had ini-
appropriate memory measure (collected 1 week later). Using tially reported if they had been relatively distressed.
residualized (partialed) scores to assess change avoids several
problems inherent in raw difference scores (Cohen Cohen,
1983). The predictor variables in this set of APIM analyses Discriminant analyses
were actor gender, actor and partner avoidance and anxiety, and We reconducted all of the reported analyses to statistically
actor and partner observer-rated distress. We also included the control for each individual’s (each actor’s) scores on
appropriate observer ratings of supportive or distancing behav-
ior displayed by the actor and the partner as control variables. 1.5
No significant main effects were found in these analyses.1,2 Low Actor Anxiety
Change in Perceived Self-Distancing
As predicted, avoidance and anxiety were both related to 1.0 High Actor Anxiety
changes in memory, but the effects were contingent on how
distressed individuals (actors) had been during the discus-
sions. As Figure 1 shows, for the analysis in which residual- 0.5
ized change in supportiveness was the dependent measure, an
interaction confirmed that more avoidant individuals (actors) 0
remembered being less supportive 1 week after the discussion Low Actor Distress High Actor Distress
than they reported immediately after the discussion, but only if
they had been rated by observers as being relatively distressed −0.5
during the discussion, β = –0.19, t(260) = –3.59, p .001.
This interaction also revealed that less avoidant individuals −1.0
remembered being more supportive 1 week after the discus-
sion than they reported immediately after the discussion, again
−1.5
only if they had been relatively distressed during the discus-
Fig. 2. The interaction of actor’s distress and attachment anxiety in
sion. As Figure 2 shows, for the analysis in which residualized predicting change in actor’s perceptions of self-distancing from Phase 1 to
change in distancing was the dependent measure, less anxious Phase 2. Slopes were computed for values 1 standard deviation above and 1
individuals (actors) recalled being more distant from their standard deviation below the mean on each independent variable.
Downloaded from pss.sagepub.com by guest on January 29, 2013

7. Attachment and Memory 257
neuroticism and for how much each individual (each actor) Schema-consistent memory changes were also associated
talked about the discussion task with his or her partner between with attachment anxiety. Less anxious individuals recalled
Phase 1 and Phase 2. When we did so, all of the significant being more distant in Phase 2 than they had reported initially
effects reported remained significant. The observed attach- if they had been relatively distressed during the discussion.
ment effects, therefore, are not attributable to shared variance Results for more anxious individuals displayed the opposite
with neuroticism or to how much partners talked about the pattern, and were consistent with such individuals’ irrepress-
discussion topic during the intervening week. ible desire to become closer to their partners and achieve
greater felt security; that is, in Phase 2, more anxious individu-
als remembered being less distant from (emotionally closer to)
Discussion their partners than they had reported initially if they had been
This study provides several novel and theoretically important relatively distressed during the discussion.
findings. The results for observer ratings of the discussions are In sum, the memory-change findings are consistent with
consistent with prior research showing that attachment orien- the needs, goals, concerns, and motives harbored by highly
tations are linked to actual behavior in attachment-relevant avoidant and highly anxious persons, but only if these indi-
situations (e.g., Collins Feeney, 2000; Simpson et al., 1996). viduals were relatively distressed when the memory was cre-
More avoidant individuals were rated as displaying less sup- ated. The tenor of these findings is also consistent with what
port and more distancing behavior during their discussions. B.C. Feeney and Cassidy (2003) found. The current study,
These results are consistent with studies that have documented however, differs in several important respects from their study.
what highly avoidant people do behaviorally to keep their Feeney and Cassidy, for example, did not examine the effects
attachment systems deactivated and quiescent (e.g., Miku- of distress on memory, did not study romantic partners, did not
lincer et al., 1993; Simpson et al., 1992). In addition, more assess the two dimensions of adult attachment (anxiety and
anxious individuals were rated as less supportive when their avoidance), and did not test individuals’ memory changes for
partners were more distressed. When their partners are dis- their own behavior in relation to the two types of insecurity
tressed, highly anxious persons may want to provide high- (anxiety and avoidance).
quality support, but they should find it difficult to do so These findings have several important implications. First,
because stressful events—especially interpersonal ones—­ they shed light on how working models may buffer and stabi-
cognitively overload anxious persons (Mikulincer Shaver, lize attachment orientations across time. Attachment orienta-
2003), reducing their ability to orient toward and effectively tions remain stable, at least in part, because of schema-consistent
soothe their partners. memory storage and retrieval processes. These findings also
Memory changes were tested by partialing each individu- help explain why it is difficult for people—even those in
al’s Phase 1 score from his or her Phase 2 score on each focal therapy—to change their attachment orientations. Even if
measure (support and distancing). The change findings, individuals want to disconfirm and alter their insecure work-
therefore, reflect memory changes after encoding, once ing models, their models are likely to twist their memories in
working models had already started to twist perceptions. model-congruent ways, particularly in stressful attachment-
Results for avoidant individuals were consistent with their relevant situations.
working models, as more avoidant individuals remembered Second, our findings indicate that attachment-based mem-
being less supportive in Phase 2 than they had reported ini- ory distortions depend on the degree to which individuals are
tially if they had been relatively distressed during the discus- distressed when memories are initially formed. These findings
sion a week earlier. Less avoidant persons, in contrast, are consistent with a core principle of attachment theory,
recalled being more supportive in Phase 2 than they had namely, that working models ought to affect how individuals
reported initially if they had been relatively distressed during think, feel, and behave most strongly when the attachment sys-
the discussion. These findings are consistent with the needs tem is activated (Bowlby, 1969, 1973; Simpson Rholes,
and goals of highly avoidant people, who yearn to limit inti- 1994). They are also consistent with a recent review of chil-
macy and maintain control and autonomy in their relation- dren’s memory for distressing events. Alexander et al. (2002)
ships. They are also in line with the needs and goals of less found that attachment models affect the type and amount of
avoidant (i.e., more secure) persons, who strive to increase information young children encode, store, and retrieve primar-
intimacy in their relationships (Mikulincer, 1998). These ily when children experience stressful events. Much as we
findings also mesh nicely with earlier behavioral observation found, children typically remember stressful events in a man-
studies of avoidance. For example, we (Simpson et al., 1992) ner that is aligned with their attachment schemas (working
found that when romantic partners are waiting to do a stress- models).
ful activity, less avoidant people offer more support if their In conclusion, what individuals respond to in relationships
partners appear distressed, whereas more avoidant people is not what they actually said or did during an interaction with
offer less support if their partners appear distressed—even their partner; rather, what they respond to is memories of the
though this is precisely the situation in which partners need interaction filtered through their working models. The current
support the most. findings highlight the important role of working models and
Downloaded from pss.sagepub.com by guest on January 29, 2013

8. 258 Simpson et al.
the motivations that underlie them in generating a “model- themes. In W.S. Rholes J.A. Simpson (Eds.), Adult attachment:
consistent reality” of a relationship. Theory, research, and clinical implications (pp. 196–239). New
York: Guilford Press.
Declaration of Conflicting Interests Cutrona, C.E. (1989). Ratings of social support by adolescents and
The authors declared that they had no conflicts of interests with adult informants: Degree of correspondence and prediction of
respect to their authorship and/or the publication of this article. depressive symptoms. Journal of Personality and Social Psy-
chology, 57, 723–730.
Funding Feeney, B.C., Cassidy, J. (2003). Reconstructive memory related
This research was supported by National Institute of Mental Health to adolescent-parent conflict interactions: The influence of
Grant MH49599 to Jeffry A. Simpson and W. Steven Rholes. attachment-related representations on immediate perceptions and
changes in perceptions over time. Journal of Personality and
Notes Social Psychology, 85, 945–955.
1. Partner variables were included in these analyses so we could per- Feeney, J.A. (1999). Adult romantic attachments and couple relation-
form better and more precise tests of the hypothesized actor effects. ships. In J. Cassidy P.R. Shaver (Eds.), Handbook of attach-
2. Gender was a significant predictor in two interactions. Even though ment: Theory, research, and clinical applications (pp. 355–377).
highly avoidant individuals recalled being more emotionally distant New York: Guilford Press.
in Phase 2 relative to Phase 1, this effect was smaller for women than Fraley, R.C., Brumbaugh, C.C. (2004). Predictors of change in
for men, β = 0.11, t(260) = 3.18, p .003. Also, women recalled being attachment security. In W.S. Rholes J.A. Simpson (Eds.), Adult
more distant in Phase 2 if their partners were more avoidant, whereas attachment: Theory, research, and clinical implications (pp. 133–
men recalled being less distant if their partners were more avoidant, 156). New York: Guilford Press.
β = –0.07, t(260) = –1.97, p .05. Fraley, R.C., Garner, J.P., Shaver, P.R. (2000). Adult attachment
and the defensive regulation of attention and memory: Examin-
References ing the role of preemptive and postemptive processes. Journal of
Alexander, K.W., Quas, J.A., Goodman, G.S. (2002). Theoreti- Personality and Social Psychology, 79, 1–11.
cal advances in understanding children’s memory for distress- John, O.P., Donahue, E.M., Kentle, R.L. (1991). The Big Five
ing events: The role of attachment. Developmental Review, 22, Inventory—Versions 4a and 54. Berkeley: University of Califor-
490–519. nia, Berkeley, Institute for Personality and Social Research.
Belsky, J., Spritz, B., Crnic, K. (1996). Infant attachment security Kashy, D.A., Kenny, D.A. (2000). The analysis of data from dyads
and affective-cognitive information processing at age 3. Psycho- and groups. In H.T. Reis C.M. Judd (Eds.), Handbook of
logical Science, 7, 111–114. research methods in social psychology (pp. 451–477). New York:
Bowlby, J. (1969). Attachment and loss: Attachment. New York: Cambridge University Press.
Basic Books. Kenny, D.A. (1996). Models of interdependence in dyadic
Bowlby, J. (1973). Attachment and loss: Separation, anxiety, and research. Journal of Social and Personal Relationships, 13,
anger. New York: Basic Books. 279–294.
Bowlby, J. (1980). Attachment and loss: Sadness and depression. Kirsh, S.J., Cassidy, J. (1997). Preschoolers’ attention to and mem-
New York: Basic Books. ory for attachment-relevant information. Child Development, 68,
Brennan, K.A., Clark, C.L., Shaver, P.R. (1998). Self-report mea- 1143–1153.
surement of adult attachment: An integrative overview. In J.A. Kobak, R.R., Duemmler, S. (1994). Attachment and conversation:
Simpson W.S. Rholes (Eds.), Attachment theory and close Toward a disclosure analysis of adolescent and adult security. In
relationships (pp. 46–76). New York: Guilford Press. K. Bartholomew D. Perlman (Eds.), Attachment processes in
Brennan, K.A., Shaver, P.R. (1995). Dimensions of adult attach- adulthood (pp. 121–149). London: Jessica Kingsley.
ment, affect regulation, and romantic relationship functioning. Mikulincer, M. (1998). Attachment working models and the sense
Personality and Social Psychology Bulletin, 21, 267–283. of trust: An exploration of interaction goals and affect regu-
Cohen, J., Cohen, P. (1983). Applied multiple regression/correla- lation. Journal of Personality and Social Psychology, 74,
tion analysis for the behavioral sciences. Hillsdale, NJ: Erlbaum. 1209–1224.
Collins, N.L., Feeney, B.C. (2000). A safe haven: An attachment Mikulincer, M., Arad, D. (1999). Attachment working models and
theory perspective on support seeking and caregiving in intimate cognitive openness in close relationships: A test of chronic and
relationships. Journal of Personality and Social Psychology, 78, temporary accessibility effects. Journal of Personality and Social
1053–1073. Psychology, 77, 710–725.
Collins, N.L., Feeney, B.C. (2004). Working models of attachment Mikulincer, M., Florian, V., Weller, A. (1993). Attachment orienta-
affect perceptions of social support: Evidence from experimental tions, coping strategies, and posttraumatic psychological distress:
and observational studies. Journal of Personality and Social Psy- The impact of the Gulf War in Israel. Journal of Personality and
chology, 87, 363–383. Social Psychology, 64, 817–826.
Collins, N.L., Guichard, A.C., Ford, M.B., Feeney, B.C. (2004). Mikulincer, M., Orbach, I. (1995). Attachment orientations and
Working models of attachment: New developments and emerging repressive defensiveness: The accessibility and architecture of
Downloaded from pss.sagepub.com by guest on January 29, 2013

9. Attachment and Memory 259
affective memories. Journal of Personality and Social Psychol- Rudolph, K.D., Hammen, C., Burge, D. (1997). Cognitive repre-
ogy, 68, 917–925. sentations of self, family, and peers in school-age children: Links
Mikulincer, M., Shaver, P.R. (2003). The attachment behavioral with social competence and sociometric status. Child Develop-
system in adulthood: Activation, psychodynamics, and interper- ment, 66, 1385–1402.
sonal processes. In M.P. Zanna (Ed.), Advances in experimental Simpson, J.A., Rholes, W.S. (1994). Stress and secure
social psychology (Vol. 35, pp. 53–152). New York: Academic base relationships in adulthood. In K. Bartholomew
Press. D. Perlman (Eds.), Advances in personal relationships: Vol. 5.
Miller, J.B. (1999). Attachment orientation and memory for attachment- Attachment processes in adulthood (pp. 181–204). London:
related events. Journal of Social and Personal Relationships, 16, Kingsley.
773–801. Simpson, J.A., Rholes, W.S., Nelligan, J.S. (1992). Support-seeking
Miller, J.B., Noirot, M. (1999). Attachment memories, models and and support-giving within couples in an anxiety-provoking situ-
information processing. Journal of Social and Personal Relation- ation: The role of attachment styles. Journal of Personality and
ships, 16, 147–173. Social Psychology, 62, 434–446.
Pietromonaco, P.R., Barrett, L.F. (1997). Working models of Simpson, J.A., Rholes, W.S., Phillips, D. (1996). Conflict in close
attachment and daily social interactions. Journal of Personality relationships: An attachment perspective. Journal of Personality
and Social Psychology, 73, 1409–1423. and Social Psychology, 71, 899–914.
Downloaded from pss.sagepub.com by guest on January 29, 2013